Phone Harness
ShawnPana/phone-harness
Control the user's phone — an iPhone through the Mac's iPhone Mirroring window, an Android over adb, or a rented cloud Android: open apps, tap, type, swipe, read the screen.
A skill your agent uses to drive an Android device from the host shell — inject taps, swipes, text, key events, and drag-and-drop via adb shell input; resize the display with wm size / wm density…
$ npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install skydoves/android-testing-skills injecting-input-and-state --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/skydoves/android-testing-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/adb/control/injecting-input-and-state .claude/skills/injecting-input-and-state && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "injecting-input-and-state" agent skill from https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state into .claude/skills/injecting-input-and-state/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "injecting-input-and-state", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-stateType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install skydoves/android-testing-skills injecting-input-and-state --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/skydoves/android-testing-skills.git skills-src && mkdir -p .agents/skills && cp -r skills-src/adb/control/injecting-input-and-state .agents/skills/injecting-input-and-state && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "injecting-input-and-state" agent skill from https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state into .agents/skills/injecting-input-and-state/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "injecting-input-and-state", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install skydoves/android-testing-skills injecting-input-and-state --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/skydoves/android-testing-skills.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/adb/control/injecting-input-and-state .cursor/skills/injecting-input-and-state && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "injecting-input-and-state" agent skill from https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state into .cursor/skills/injecting-input-and-state/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "injecting-input-and-state", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/skydoves/android-testing-skills.git --path adb/control/injecting-input-and-state--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install skydoves/android-testing-skills injecting-input-and-state --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/skydoves/android-testing-skills.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/adb/control/injecting-input-and-state .gemini/skills/injecting-input-and-state && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "injecting-input-and-state" agent skill from https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state into .gemini/skills/injecting-input-and-state/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "injecting-input-and-state", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install skydoves/android-testing-skills injecting-input-and-stateInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/skydoves/android-testing-skills.git skills-src && mkdir -p .github/skills && cp -r skills-src/adb/control/injecting-input-and-state .github/skills/injecting-input-and-state && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "injecting-input-and-state" agent skill from https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state into .github/skills/injecting-input-and-state/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "injecting-input-and-state", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install skydoves/android-testing-skills injecting-input-and-state --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/skydoves/android-testing-skills.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/adb/control/injecting-input-and-state .opencode/skills/injecting-input-and-state && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "injecting-input-and-state" agent skill from https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state into .opencode/skills/injecting-input-and-state/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "injecting-input-and-state", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
injecting-input-and-stateA skill your agent uses to drive an Android device from the host shell — inject taps, swipes, text, key events, and drag-and-drop via adb shell input; resize the display with wm size / wm density…
Injecting Input And State is an agent skill from skydoves/android-testing-skills. Use this skill to drive an Android device from the host shell — inject taps, swipes, text, key events, and drag-and-drop via adb shell input; resize the display with wm size / wm density; flip hermetic-test settings with settings put global windowanimationscale 0; reset app state with pm clear vs am force-stop; launch Activities with am start -n pkg/.Activity; and use the modern cmd <service wrapper (cmd wifi, cmd connectivity, cmd uimode night, cmd locale set-app-locales) instead of the deprecated svc aliases on…
Its SKILL.md is about 4.8k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.
It sits in Mobile, covering Mobile testing and debugging. It works with Android. The repository describes itself as: ⚡️ A set of skills for Android testing: Compose UI, AndroidX Test, JVM unit tests, and ADB. The licence is Apache-2.0.
Read from SKILL.md and the folder at commit 8665ed5. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Shell commands in SKILL.md call:
adbFrom the folder's file list and the shell code blocks in SKILL.md.
Links to these hosts (documentation or services it may open):
developer.android.comFrom URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Injecting Input And State loads about 4.8k tokens when it runs. Until then it costs about 213 tokens; SKILL.md has 1,570 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
The full file from skydoves/android-testing-skills at commit 8665ed5, republished under its Apache-2.0 licence (© skydoves). 1,570 words, ~4,825 tokens.
.claude/skills/injecting-input-and-state/SKILL.md (or your agent's skills folder).This skill covers the on-device shell surface a test/QA engineer drives over adb to simulate user gestures, override the screen configuration, flip system settings, reset app state, and launch Activities. Transport (adb devices, USB/TCP) lives in ../../architecture/understanding-adb-architecture/SKILL.md; capturing artefacts after driving the device lives in ../../capture/capturing-screenshots-and-screenrecord/SKILL.md.
wm size 411x731, wm density 320) without recreating an emulator AVD.window_animation_scale = 0 etc.).am force-stop is enough (it is not).adb shell svc wifi disable and silently does nothing on API 30+ — the modern cmd wifi set-wifi-enabled disabled is the fix.--ez, --ei, --es).../../capture/capturing-screenshots-and-screenrecord/SKILL.md, ../../observability/extracting-logs-with-logcat/SKILL.md, ../../transfer/extracting-test-artifacts/SKILL.md.../../../instrumentation/scenarios/launching-activities-with-activityscenario/SKILL.md or UiAutomator.../../apps/installing-and-managing-apps/SKILL.md.../../devices/connecting-to-devices/SKILL.md or ../../devices/connecting-over-wifi/SKILL.md.adb get-state returns device).adb shell is double-shell (ssh-style) since Platform Tools 23 — load-bearing for any string with spaces (developer.android.com/tools/adb#shellcommands).cmd wrapper APIs and reliable shell exit codes: device API 24+ (cmd introduced API 24).input — synthesised user eventsinput is the on-device wrapper around InputManager.injectInputEvent(). The full grammar:
input [<source>] <command> [<arg>...]<source> defaults per command (touchscreen for tap/swipe, keyboard for text/keyevent).
adb shell input tap <x> <y>
adb shell input swipe <x1> <y1> <x2> <y2> [<duration_ms>]
adb shell input draganddrop <x1> <y1> <x2> <y2> [<duration_ms>]
adb shell input roll <dx> <dy> # rotary/trackball axes
adb shell input motionevent <DOWN|UP|MOVE|CANCEL> <x> <y>swipe defaults to ~300 ms when <duration_ms> is omitted. To synthesise a long-press, swipe in place for the long-press timeout: input swipe <x> <y> <x> <y> 1000. draganddrop differs from swipe in that it emits the drag-and-drop gesture sequence (DOWN → MOVE → UP) consumable by View.OnDragListener, not the same event stream as swipe.
adb shell input text "<string>"Caveats (from developer.android.com/tools/adb):
%s. input text "hello world" types hello, then nothing.', ", &, (, ), <, >, |, ;, \, backtick) MUST be escaped twice — once for the local shell, once for the device-side shell.input text. Use UiAutomator UiObject2.setText(...) for those.adb shell input keyevent [--longpress|--doubletap] <KEYCODE> [<KEYCODE>...]
adb shell input keyevent <NUMERIC_CODE>Test-relevant KEYCODE_* (KeyEvent constants from frameworks/base; full table in research note A2):
| Mnemonic | Numeric | Notes |
|---|---|---|
KEYCODE_HOME | 3 | Press Home (does not always wake screen). |
KEYCODE_BACK | 4 | Back button. |
KEYCODE_DPAD_UP/DOWN/LEFT/RIGHT/CENTER | 19-23 | TV / focus traversal. |
KEYCODE_VOLUME_UP/DOWN/MUTE | 24/25/164 | |
KEYCODE_POWER | 26 | Toggles screen off/on. |
KEYCODE_TAB | 61 | |
KEYCODE_ENTER | 66 | |
KEYCODE_DEL | 67 | Backspace. |
KEYCODE_MENU | 82 | |
KEYCODE_APP_SWITCH | 187 | Recents button. |
KEYCODE_WAKEUP | 224 | Wake without unlocking. |
KEYCODE_SLEEP | 223 | Force sleep. |
Wake + dismiss-keyguard for emulator with no lock:
adb shell input keyevent KEYCODE_WAKEUP
adb shell input keyevent 82 # menu, dismiss the AOSP swipe-upinput <source> <command> maps to InputDevice.SOURCE_*. Use a source prefix when a screen has multiple input listeners or when targeting non-touch hardware:
| Source | Typical use |
|---|---|
touchscreen | Default for tap, swipe, draganddrop. |
touchpad | External touchpads (Chromebook / Android tablet keyboard). |
dpad | TV/focus traversal: input dpad keyevent KEYCODE_DPAD_DOWN. |
keyboard | Default for text, keyevent. |
mouse | Mouse buttons + scroll. |
trackball | Trackball roll. |
gamepad / joystick | Gamepad buttons / joystick axes. |
stylus | Pressure-aware touch (S-Pen). |
wm — display size and density overrideswm is the shell wrapper for IWindowManager. The test-relevant subcommands:
adb shell wm size # read current logical size
adb shell wm size 1080x1920 # override
adb shell wm size reset # restore device default
adb shell wm density # read current dpi
adb shell wm density 320 # override
adb shell wm density reset
adb shell wm dismiss-keyguard # insecure keyguard only
adb shell wm user-rotation [free|lock [<rotation>]] # 0,1,2,3 = 0/90/180/270Use case — screen-size / density matrix without recreating the emulator:
# Per configuration in the matrix
adb shell wm size 411x731
adb shell wm density 240
# … run tests …
adb shell wm size reset
adb shell wm density resetCaveats:
reset in @AfterClass / suite teardown — otherwise the device stays in the modified configuration after the run, breaking later runs and any human user.wm size / wm density trigger a configuration change that recreates the foreground Activity. Treat them as between-test knobs, not mid-test.am display-size / am display-density are equivalent legacy spellings; wm is preferred.settings — hermetic test stateadb shell settings [--user <id>] get <namespace> <key>
adb shell settings [--user <id>] put <namespace> <key> <value> [<tag>] [default]
adb shell settings [--user <id>] delete <namespace> <key>
adb shell settings [--user <id>] reset <namespace> [<package>] [untrusted_defaults|untrusted_clear|trusted_defaults]
adb shell settings [--user <id>] list <namespace><namespace> is one of global, system, secure.
Set all three to 0 for stable instrumented and screenshot tests:
adb shell settings put global window_animation_scale 0
adb shell settings put global transition_animation_scale 0
adb shell settings put global animator_duration_scale 0Or via Gradle: testOptions.animationsDisabled = true.
| Namespace | Key | Purpose |
|---|---|---|
system | font_scale | Float (e.g. 1.0, 1.3); type-scale tests. |
secure | accessibility_enabled | 1 to enable a11y; pair with enabled_accessibility_services. |
secure | enabled_accessibility_services | CSV pkg/.Service for a11y tests. |
secure | default_input_method | Force a known IME. |
secure | show_ime_with_hard_keyboard | 1 to keep IME visible on emulators. |
secure | long_press_timeout | Stabilise long-press tests. |
global | airplane_mode_on | Pair with am broadcast android.intent.action.AIRPLANE_MODE. |
global | policy_control | E.g. immersive.full=* to hide system bars. |
am force-stop vs pm clear| Command | Kills processes? | Clears data/cache/prefs? | Resets runtime permissions? |
|---|---|---|---|
am force-stop <pkg> | Yes | No | No |
pm clear <pkg> | Yes (implicitly) | Yes | Yes |
cmd activity stop-app <pkg> | Yes (modern, API 28+) | No | No |
For a hermetic between-test reset, use pm clear. For "kill the process so the next launch is cold", am force-stop is enough. See ../../apps/installing-and-managing-apps/SKILL.md for the full pm surface.
am start# Most reliable for tests: explicit component.
adb shell am start -n com.example.app/.MainActivity
# Force-stop first, then launch (clean cold start).
adb shell am start -S -W -n com.example.app/.MainActivity
# Launch by action + URI (default handler for https://).
adb shell am start -a android.intent.action.VIEW -d 'https://example.com'
# Pass typed extras.
adb shell am start -n com.example.app/.MainActivity \
--es screen "settings" --ez debug true --ei count 42
# Return to launcher between tests.
adb shell am start -a android.intent.action.MAIN -c android.intent.category.HOMEIntent specifier flags (abridged from developer.android.com/tools/adb#IntentSpec):
| Token | Effect |
|---|---|
-a action | Intent action (e.g. android.intent.action.VIEW). |
-d data_uri | Data URI. |
-t mime_type | MIME type. |
-c category | Category. |
-n component | Explicit component (pkg/.Activity). |
-f flags | Hex setFlags() value. |
| `--es | -e key value` |
--ez key bool | Boolean extra. |
--ei key int | Int extra. |
--el key long | Long extra. |
--ef key float | Float extra. |
--eu key uri | URI extra. |
--ecn key component | ComponentName extra. |
--eia/--ela/--efa key v[,v...] | Array extras. |
--activity-clear-top etc. | Map to FLAG_ACTIVITY_*. |
-S | Force-stop the target before starting. |
-W | Wait for launch to complete. |
cmd <service> — the modern wrappercmd is a generic wrapper that calls into a binder service's onShellCommand. Available since API 24. To list services: adb shell cmd -l (or dumpsys -l). Test-relevant entry points:
| Invocation | Effect |
|---|---|
cmd activity stop-app <pkg> | Modern equivalent of am force-stop. |
cmd activity broadcast … | Same as am broadcast. |
cmd package install -r -t -g <path> | pm install equivalent. |
cmd package compile -m speed -f <pkg> | Force AOT compilation. |
`cmd appops set <pkg> <op> allow | deny |
cmd jobscheduler run -f <pkg> <jobId> | Force a JobService to run now. |
| `cmd uimode night yes | no |
cmd locale set-app-locales <pkg> --locales en-US | Per-app locales (Android 13+). |
cmd statusbar expand-notifications / collapse | Expand/collapse the shade. |
| `cmd wifi set-wifi-enabled enabled | disabled` |
| `cmd connectivity airplane-mode enable | disable` |
cmd testharness enable | Reset device to a hermetic test-harness state (Android 10+). |
svc deprecation on API 30+svc data, svc wifi, svc bluetooth are no-ops on API 30+ unless invoked from a privileged shell (corpus §I.10). The preferred replacements are the cmd wrappers above.
svc wifi disable on API 30+# WRONG
adb shell svc wifi disable
# WRONG because: silently no-ops on API 30+ since the svc CLI was demoted from a privileged
# helper to an ordinary shell command. Exit code 0, Wi-Fi still on. The script proceeds
# under a wrong assumption — exactly the worst kind of failure mode.# RIGHT
adb shell cmd wifi set-wifi-enabled disabled
adb shell cmd wifi status # verifyam force-stop for a between-test reset# WRONG
adb shell am force-stop com.example.app
./run-next-test.sh
# WRONG because: am force-stop kills the process but leaves shared prefs, databases, the
# image cache, and granted runtime permissions intact. The "next test" inherits state
# from the previous run.# RIGHT
adb shell pm clear com.example.app
adb shell pm clear com.example.app.test # also reset test process state
./run-next-test.sh(Test Orchestrator with clearPackageData: 'true' is the equivalent inside Gradle — see ../../tests/running-instrumented-tests-via-adb/SKILL.md.)
input text "hello world"# WRONG
adb shell input text "hello world"
# WRONG because: spaces in `input text` MUST be encoded as %s. The above types only "hello".# RIGHT
adb shell input text 'hello%sworld'adb shell settings put global window_animation_scale 0
adb shell settings put global transition_animation_scale 0
adb shell settings put global animator_duration_scale 0
adb shell pm clear com.example.app
adb shell am start -S -W -n com.example.app/.MainActivity-S force-stops first; -W blocks until the launch completes.
adb shell input dpad keyevent KEYCODE_DPAD_DOWN
adb shell input dpad keyevent KEYCODE_DPAD_DOWN
adb shell input dpad keyevent KEYCODE_DPAD_CENTERSource-prefixing dpad ensures the events route to TV-style focus handlers, not the touchscreen pipeline.
cmd wifi set-wifi-enabled / cmd connectivity airplane-mode / cmd bluetooth_manager enable on API 30+. MUST NOT rely on svc wifi / svc bluetooth / svc data — they no-op silently on modern releases.pm clear <pkg> (not am force-stop) when the goal is a hermetic between-test reset. force-stop only kills processes; data is untouched.input text as %s; MUST double-shell quote (ssh-style) any string with shell metacharacters. Platform Tools 23+ (developer.android.com/tools/adb#shellcommands).wm size reset and wm density reset in suite teardown when the suite overrode them. A leaked override breaks every subsequent run on that device.0 (or use testOptions.animationsDisabled = true in Gradle) before any UI-asserting test run.Thread.sleep between adb shell input calls to "let the UI catch up". Inside an instrumentation test, use Compose's idle synchronisation (../../../compose/synchronization/synchronizing-with-idle/SKILL.md); from a host script, use adb shell am start -W to block on Activity launch and adb shell wait-for-device between reboots.adb shell am instrument returns a meaningful exit code without -w. See ../../tests/running-instrumented-tests-via-adb/SKILL.md and ../../automation/scripting-adb-for-ci/SKILL.md.am start -S -W -n pkg/.Activity for cold-start launches in tests — -S resets, -W blocks until ready.adb shell settings get global window_animation_scale returns 0 (or 0.0) before tests run; same for transition_animation_scale and animator_duration_scale.scripts/ or .github/workflows/ uses svc wifi / svc bluetooth / svc data.am force-stop between tests is paired with (or replaced by) pm clear.input text "<string with spaces>" exists; all spaces are %s.wm size <override> in setup has a matching wm size reset in teardown.am, pm, input, wm, settings, intent specs (https://developer.android.com/tools/adb).adb shell since Platform Tools 23.KEYCODE_* constants table.testOptions.animationsDisabled.tasks/research/A2-adb-shell-commands.md — full input / wm / settings / am / pm / cmd / svc tables and verbatim quotes.../../architecture/understanding-adb-architecture/SKILL.md — three-piece architecture, server, daemon.../../devices/connecting-to-devices/SKILL.md — adb devices states, wait-for-device.../../devices/connecting-over-wifi/SKILL.md — adb pair / adb connect.../../apps/installing-and-managing-apps/SKILL.md — full pm install / pm uninstall / pm grant surface.../../tests/running-instrumented-tests-via-adb/SKILL.md — am instrument -w -r.../../capture/capturing-screenshots-and-screenrecord/SKILL.md — capturing artefacts after driving the device.../../observability/extracting-logs-with-logcat/SKILL.md — reading device logs.../../transfer/extracting-test-artifacts/SKILL.md — adb pull / adb push / run-as.../../automation/scripting-adb-for-ci/SKILL.md — bash idioms, retries, port forwarding.../../../instrumentation/scenarios/launching-activities-with-activityscenario/SKILL.md — launching from inside an instrumentation test.../../../fundamentals/strategies/applying-testing-strategies/SKILL.md — when to drive via adb vs instrumentation.© skydoves, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Just SKILL.md in adb/control/injecting-input-and-state of skydoves/android-testing-skills.
Open the folder on GitHubat commit 8665ed5
Injecting Input And State next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Injecting Input And State this skillskydoves/android-testing-skills | 333 | — | ~4.8k | Automated safety check: Pass | Apache-2.0 | |
| Phone HarnessShawnPana/phone-harness | 3.2k | — | ~4.3k | Automated safety check: Pass | MIT | |
| Mobile QAtloncorp/tlon-apps | 107 | — | ~2.4k | Automated safety check: Pass | MIT | |
| Androidyang1ming/android-harness | 176 | — | ~259 | Automated safety check: Pass | MIT | |
| Debug Receiverstimusus/Shuttle2 | 229 | — | ~1.9k | Automated safety check: Pass | Apache-2.0 | |
| Debug Bridgegetknit/knit | 132 | — | ~2.2k | Automated safety check: Pass | GPL-3.0 |
ShawnPana/phone-harness
Control the user's phone — an iPhone through the Mac's iPhone Mirroring window, an Android over adb, or a rented cloud Android: open apps, tap, type, swipe, read the screen.
tloncorp/tlon-apps
Run a mobile QA checklist on a physical Android device over adb for tlon-apps, then triage what fails into fixes.
yang1ming/android-harness
Direct Android device control through ADB. An agent skill from yang1ming/android-harness.
timusus/Shuttle2
Drive the S2 debug build's playback and queue over ADB broadcasts — play the whole library, play/pause, skip, seek, remove a queue item, toggle shuffle/repeat, dump playback state as JSON, reimport…
getknit/knit
Drive and verify Knit on a device or emulator through the headless debug bridge (am broadcast to app.getknit.knit.debug.<ACTION, replies as JSON) — send a message on one phone and confirm it landed…
Moustachauve/WLED-Android
Production-ready scripts for Android app testing, building, and automation.
skydoves/android-testing-skills
A skill your agent uses to verify Compose layout measurements from a UI test using assertWidthIsEqualTo, assertHeightIsEqualTo, assertWidthIsAtLeast, assertHeightIsAtLeast…
skydoves/android-testing-skills
A skill your agent uses to verify a Compose semantics node's properties from a UI test using assertExists, assertDoesNotExist, assertIsDisplayed, assertIsNotDisplayed, assertIsDeactivated…
skydoves/android-testing-skills
A skill your agent uses to render every Jetpack Compose @Preview as a screenshot on a real Android device or emulator and publish a browsable HTML catalog from CI.
skydoves/android-testing-skills
A skill your agent uses to capture visual artefacts from a device for test failures, golden image generation, QA repro, and demo videos.
skydoves/android-testing-skills
A skill your agent uses to pick the correct Compose UI test entry point.
skydoves/android-testing-skills
A skill your agent uses to pick which behaviors to cover in an Android test suite using Google's five-category state vocabulary plus the explicit "what NOT to test" list from…
Works with
Categories
A skill your agent uses to drive an Android device from the host shell — inject taps, swipes, text, key events, and drag-and-drop via adb shell input; resize the display with wm size / wm density…. Injecting Input And State is an agent skill from skydoves/android-testing-skills.
Injecting Input And State fits situations like: drive an Android device from the host shell — inject taps; drag-and-drop via adb shell input; resize the display with wm size / wm density; flip hermetic-test settings with settings put global windowanimationscale 0.
Run `npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a claude-code`. Or copy the skill folder (adb/control/injecting-input-and-state in skydoves/android-testing-skills) into .claude/skills/injecting-input-and-state in your project. Claude Code loads it when a task matches its description.
Run `npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a codex`. Or copy the skill folder (adb/control/injecting-input-and-state in skydoves/android-testing-skills) into .agents/skills/injecting-input-and-state in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add skydoves/android-testing-skills --skill injecting-input-and-state -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/injecting-input-and-state, .gemini/skills/injecting-input-and-state, .github/skills/injecting-input-and-state and .opencode/skills/injecting-input-and-state in your project.
Going by SKILL.md and its folder, Injecting Input And State needs the command-line tools its instructions call (adb).
SKILL.md names 1 domain. As links in the text: developer.android.com. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
Injecting Input And State is published under the Apache-2.0 licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.
About 4.8k tokens (SKILL.md is roughly 19k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
Skills that share tags, products or a category with Injecting Input And State: Phone Harness (ShawnPana/phone-harness, 3.2k stars), Mobile QA (tloncorp/tlon-apps, 107 stars), Android (yang1ming/android-harness, 176 stars) and Debug Receivers (timusus/Shuttle2, 229 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
skydoves (a GitHub user) maintains it in skydoves/android-testing-skills, which has 333 GitHub stars. The repository holds 50 skills in this directory. The repository was last updated on May 25, 2026.
Source: skydoves/android-testing-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.